The Porsche 911 and the BMW boxer motorcycle show what can be achieved by consistently improving and refining an architecture.  The strong points have been improved, the weak parts eliminated untill the value of package is just magnificent. Just follow the learning curve..

Lately I started to wonder if this same principle would apply to FttH deployment.

The Netherlands is a good reference because the basic architecture for FttH has been the same for the last 5 years or more. The dominant (> 90%) architecture deployed is point-to-point with 2 fibers per home, and the volume has grown to more than 350.000 homes passed per year. And last but not least, it is possible to get some (mostly off-the-record)  intelligence on what is really happening.

The big picture substantiates the idea that practice trumps theory. The construction companies have managed to cut the average cost (Capex)  per connection by approx. 15 % in the last 5+ years. That is already impressive, and there is more.  The pressure to reduce Capex has lead to various experiments with new materials (packaging) and new processes that are very promising in my opinion. (The details are not yet available for publication). Some of these innovations  may turn out to be generic and applicable in other countries as well, some of them may be specific for the Dutch urban areas.

This observation shows that generic cost calculations made by consultants may seriously be overestimating the actual investment levels, as no learning curve effect is taken into account.  It shows also that sticking to an architecture and keep on improving and improving and improving pays off very well. Which is good news for FttH…..

None.

In some homes the reach and performance of wifi is laughable. Concrete walls with steel reinforcement, or floor radiation heating tubes filled with water are very effective cages of Faraday. Or the opposite is true: an abundance of access points, interfering with each other or oversteering the wireless receiver.

Wireless tethering your devices is great, but you absolutely do need a good wired network to most rooms in the home to get the performance fiber can give you.

Which never has been a point of consideration for architects or builders in the past. At best some tubes for PSTN/voice cables and some tubes for coax cables ( TV) to 2 or 3 rooms were planned and built during construction. Ok, if you have been educated in the 60’’s and 70’s and have designed homes in the 80’s and early 90’s, you are excused for not foreseeing the rise of the Internet and the importance of connectivity. So in these homes you have to add on the home network which can be a nuisance and a lot of work.

It is a bloody shame however if an architect fails to take this into account in this century, designing homes as if this revolution has not taken place and connectivity has become as important an utility as electricity. Yet most architects still do.

My recent experience has been in Amsterdam with new homes, built in 2010. Very nice family homes, targeted at middle upper class and higher with kids, multistory, square in the middle of a borough claiming to be very modern. No tubes for network cables, wifi reach is dismal , only a couple of connection points for TV and for voice.  WTF??

Even worse: if you try to reuse existing tubes for voice cabling or electricity cabling, adding data cables (Plastic Optical Fibre can be put in the same duct as electricity) to various rooms, it is a nightmare. To save money flexible PVC tubes have been used.

The drawback of these tubes is the high friction when you want to pull wires through them, making it impossible in practice to add even a tiny cable in an existing tube.

So in desperation the home owner of this expensive new house is drilling holes through walls and mounting plastic cable ducts along walls and floors to get the connectivity he needs. Ugly but there is no other option. Something that could have been integrated beautifully and for a pittance if the architects would have lived in this century, not the previous one.

We have still a long way to go….

The industry wants to keep the dichotomy alive, as the difference of  business models and control parameters generates income. Recently an Ericcson executive predicted publicly that soon there will be 5 billion mobile broadband connections, far outstripping the fixed lines (currently less than 1 billion).

The reality you can observe everyday in the developed sections of the world is that most connectivity is consumed semi-static. Wireless yes, but not mobile in the sense of moving all the time.

It is great to have a high performance wireless tether in the location where you work, sit, talk and enjoy yourself. Wifi fits the bill.  Roaming aka walking from one spot to the next and expecting to stay connected happens every now and then. This slow roaming can easily be supported if you do not mind small hiccups. (Eduroam shows how, with a single-sign-on and authentication worldwide (!!)).

It would be absolutely silly financially to try to provide connectivity for this purpose through the traditional mobile networks structure, even with LTE. There is no way there is enough bandwidth available unless you replicate the density of wifi in the home: you need many very local wireless cells.

It may take a while before this reality hits the boardrooms, but we need to start thinking of (1) fixed lines to the everywhere (2) hyperlocal wireless tethers with seamless roaming (Eduroam for business/personal use) (3) mobile broadband from large towers to fill the gaps.

The initial service offerings were middle of the road only, with hardly any distinction between VDSL and full FttH. Understandable if you only look at yourself  as a company (“coherent uniform services offering” I imagine someone in Marketing say), but not smart if you want to compete head-on with cable companies like UPC and Ziggo that do not hesitate to exploit the potential of HFC.

In the last couple of months a number of low-profile announcements and introductions have been made that show a different much more agressive attitude.

- An online PVR, allowing a user to tape up to 6 shows in parallel in KPN’s cloudbased storage-and-replay facility

- A highly appreciated Ipad IPTV app which allows you to view the IPTV services (including delay TV and the like) on your Ipad next to your TV (supported on the home wifi network only)

- A 500/500 Mbps tier, to be tested in a limited group.

Right, that is called exploiting your new potential. I wonder what the next step will be.

Take a look at what’s available…..

Our friend Kai Seim has been instrumental in  getting such a community connected to FttH.

This community  of 1290 households in 550 buildings is situated in the  hills of south-west Germany, leading up to the Swiss Alps. The broadband situation was bad: most people could not get more than 1 Mbps down, nobody would get over 3 Mbps. As a direct result of bad connectivity more and more self-employed people and small entrepeneurs were relocating , a real threat to the community.

Deploying  FttH was the answer, even when facing the difficulties of dispersed housing, differences in elevation (between 170 and  900 meters above sealevel) and large temperature differences (-20C to +30 C). The community owned network (passive and active) took a  total cost of deployment of approximately Euro 3000 per connection, not bad for such a difficult  area with low density.  Part of the investment has been subsidized by the “Bundesland”, the regional government.

Most of the cables have been buried, some strung on poles and some deployed in sewer pipes. The sewer pipe solution was not chosen for its cost-effectiveness : it was not cheaper than digging in this situation. It was much faster and created less circulation problems.

The services are supplied by Telsakom, a RWE company. As most people have satellite TV the offerings are dual play, starting with 49,90 per month for 50/2 Mbps + unlimited voice in Germany to fixed lines.

Apparently there is a viable alternative  for remote communities. It raises the question why we do not learn from the past where we have chosen (telephony network) to connect everybody and average out the differences in investment between dense and remote area’s. The benefits to the wealth of the nation were much bigger than costs, and it seems clear that the same applies again.

The old mantra of “telco must be able to get a negotiation position to force application providers like Google to pay for the use of expensive pipes” is repeated again and again. Which is not only a financially silly statement (subscribers pay enough to cover costs and profit, application providers pay already for the bandwidth and dataflow they cause, the cashflow of application providers is way too small compared to telco’s), but in other sectors this would be considered extortion. Like trying to force electrical appliance makers to pay extra to the grid owners because they cause so much electrical traffic.

In Parliament a rift is seen between the proponents of uncompromised net-neutrality provisions (nothing else is allowed) and proponents of net-neutrality with the possibility to differentiate to get a wider range of options (cheaper) for consumers. For instance for someone who does not want to use VOIP over Internet on mobile devices, only wants to use it to check mail, surf the web, do social media.

Unfortunately this has become a partisan issue which may hurt the progress of practical lawmaking that gets us forward.

So here is my solution to bridge that gap:

• Enforce that unfiltered open full net-neutral Internet access is always offered for every offered combination of speed and data cap (if applicable), by every provider
• Allow the provider to differentiate from that standard, by blocking certain types of applications (like VOIP)
• Forbid protectionist differentiation where one or a specific list of applications (brands) is blocked to keep a more profitable applications alive

The standard unabridged access may be more expensive, but the transparancy on pricing and comparison between differentiated packages (provided there is enough competition) will keep everybody in check.

The headlines  of the  Sandvine Global Internet Phenomena Report Spring 2011 have been repeated in many blogs. The focus on the size of Netflix in the USA as a percentage of (peak or average) traffic , and consequently on the question if Netflix will “swamp the Net” obscures unfortunately some  interesting information in the report.

Such as the difference in average (mean) data consumption per month between North America and the EU for fixed Internet access.

This difference supports the claim that broadband in the EU generally is faster.

The second observation is that IPTV, delivered over the same infrastructure as Internet access, generates a LOT more data traffic  than all Internet access combined.

In the Netherlands the average TV viewing time per month per person is  65 hrs, whereas in the USA it is 110 hrs according to Nielsen. If we assume a minimum per TV channel of 5 Mbps for SD quality and 12 Mbps for HD quality  this translates to  146 Gbyte/351 Gbyte of data per month (Netherlands). The total will be more, because  there are more TV sets and more persons  per household.

So it makes you wonder: why do people worry if   “Real-Time Entertainment” (as Sandvine lumps all video over the Internet together) can be sustained, while at the same time IP TV  which demands way more data is delivered over the same access  infrastructure, without any problem?  Even stronger, one could leave the TV set on 24/7, consuming Terabytes, without incurring a penalty for exceeding data caps. (!) Yet for Internet access frugality is discussed as being virtuous and economically necessary……

IPTV proves that the delivery of large amounts of data to your home is no problem at all, that the infrastructure to deliver this amount of data across the country  is available.

No, this does not apply to all video from all servers to all Internet users in the world, but that is not the point. The point is that IPTV shows that a large significant part of the existing infrastructure is already capable of delivering way more data than is discussed in the Sandvine report.

[ update: details on how IP-TV is delivered versus OTT TV]

The broadband access line to a home is shared between IP-TV, Internet access, Voice and other types of communication (broadcast TV/radio in cable or FttH for instance). The broadband access line has a fixed cost given its peak capacity and is the major cost factor (other than license costs for TV). Its cost does not depend on the amount of data that is transmitted, only on the peak capacity.

Linear IP-TV (non-Internet) usually is delivered by the telco de-duplicated to a local concentration point/exchange location, where every subscribers gets the 2 or 3 streams selected. (On-demand TV is usually delivered from a higher aggreggation point.)

Youtube or Netflix go via the Internet (OTT), they deliver their streams at a higher aggregation level to the ISP, be it an IX or a peering point. They own or hire a widely distributed Content Delivery Network (CDN) to minimize the gap between their servers and the local concentration point for the access line of the end-user. The variable cost for the ISP is in the peak transmission capacity between the CDN and the local aggregation point where the access line starts. The main issue there is that (on-demand) video traffic does not fit in the statistical multiplexing ratio’s of the past, that allowed high oversubscription ratio’s without complaints of the subscribers. This cost must either be born by the broadband access line subscriber or someone else. The costs of this middle mile transmission capacity expansion are (per subscriber) not insignificant, but not major when compared to the absolute subscription level, at least in relatively densely populated area’s. (see previous posts)

The rational thing to do from a technological and operational point of view is the have the OTT content providers to expand their CDN’s (at their cost) to deeper into the network,  while increasing middle mile capacity paid for by broadband access subscribers. There is no reason why this could not be accomodated at very acceptable costs.

The business side has a different observation: the revenue source of linear (and on-demand ) TV is potentially endangered by  the likes of Netflix, especially if the price level of a TV package is high.

The tension between these 2 perspectives is noticeable.

They got talking:  about FttH, about broadband  access speeds of 100 Mbps and 1 Gbps and so on. So what bitrates does Dutchview use for their recordings? Well, for SD 80 Mbps, for HD 120 Mbps and for material used for Bluray discs a minimum of 135 Mbps. The 135 Mbps is also demanded by the BBC for any recording. Why is this so much higher (10 times or more) than what is delivered to the home nowadays? One of the reasons is that you lose quality when editing with lower bitrates and higher compression. So the quality of the image with these higher bitrates is better? Yes, visibly so.

Right.

This digital divide is already creating tensions. The question is raised how to finance deployments beyond these islands of fiber, in the  gaps in between, and what the role of the government should be in mitigating this problem. An unanswered question up till now.

Lowering the cost of deploying fiber in less dense area’s is the other side of the equation.

In Finland cable ploughing has become the construction method of choice, where ever possible.  Ploughing is literally a tractor ploughing alongside the road while inserting a cable. The pictures show how its done.

The depth at which the cable is laid can be set between 30 and 100 cm. The costs of ploughing are between Euro 0,50 and Euro 1,50 per meter, significantly cheaper than building areal cabling where the pole right costs alone are over 1 Euro per meter.

Seems like a very viable method for countries with soil than can be ploughed and some level of organisation of underground cabling and pipes (you don’t want a plough to cut through other pipes and cables).

(hat tip Fredrik)